Seeking recommendations on statistics textbooks

[Relativism]

Hello. I'm currently taking a lower level course in statistics. It is an okay course; however, it is not very rigorous and my professor is a "this is what you need to know" sort of teacher versus one who explains the theory and reasons behind the equations as well. The required textbook for the course is similarly bad.

I have a hard time learning by rote, so I'd prefer to supplement the course material with a book that discusses statistics and probability and the theory behind them; yet, I'd also like a text that is not so esoteric as to prevent me, a lower division undergrad, from understanding without serious difficulties.

My current level of education includes the basic lower division maths: calculus, linear algebra, differential equations.

Thank you for your time.

 

[atyy]

I learned statistics from genetics - ##\chi^{2}## but how about any standard text like:
https://www.amazon.com/dp/0072472278/?tag=pfamazon01-20
https://www.amazon.com/dp/0538733527/?tag=pfamazon01-20

I think the subject is pretty standard, like first year university physics in the US, so almost any text will be the same. If you want something more mathematical, try things like:
https://www.amazon.com/dp/0321795431/?tag=pfamazon01-20
https://www.amazon.com/dp/0534243126/?tag=pfamazon01-20

A major alternative school of statistics is Bayesian statistics, for which http://www.inference.phy.cam.ac.uk/itila/book.html is absolutely terrific!

Machine learning is also a type of statistics, so one could look at http://www.cs.ubc.ca/~murphyk/MLbook/ which is really good.

Bayesian statistics is rarely truly Bayesian, and is just a convenient tool in most cases. But there is an absolutely beautiful and impractical Bayesian viewpoint that was developed by de Finetti: http://en.wikipedia.org/wiki/Coherence_(philosophical_gambling_strategy).

 

[Relativism]

atyy, thank you for the reply. I might purchase both a book on the mathematical side and on the practical side.

For now though, I think I'll stick to the primary school of statistics until I have my footing, but Bayesian will be an interesting Wikipedia read for now :)

Thank you again.

 

[jasonRF]

I just wanted to comment a little on the books mentioned. First, look in your university library before buying anything - the best way to tell if a book is useful for you is to actually look at it. Most of these books (Bevington being the only real exception) would require a lot of time to work through. I would recommend doing this after your semester is over, unless you feel like you have adequate time during the semester. Also as a caveat - I am an engineer that uses stats a lot, but am not an expert, so take my advice in that context.

Devore is often used in sophomore level courses for engineers; it is calculus based and pretty practical. If you decide to buy get a used copy of an old edition. You should be able to pick one up at amazon for a couple dollars plus shipping. I used to own the 3rd edition which was fine - I don't think this book is anything special but it isn't bad and I am not familiar with the standard competing books that you might compare it to.

I own the first edition of Bevington, which seems to be written for scientists analyzing data from lab experiments. It is a nice book for such a use, and has a reasonable discussion of parameter estimation and curve fitting. I would look for newer editions than the first which is a little old fashioned.

Hogg and Craig (intro to mathematical statistics) is a very goood book suitable for upper division students in pure and applied sciences. I really like it - it has most of what I would want in a stats book, and again recommend used copies of old editions. I think the 5th edition (the one I know the best) is excellent. I got rid of my copy and regret it!

Cassella and Berger is almost always used in graduate level courses; it emphasizes math much more than understanding. I own a copy of the first edition and really don't like it for the most part. I don't recommend jumping to that level. I would trade it for any edition of Hogg and Craig any day of the week!

good luck,

jason

 

[jasonRF]

A major alternative school of statistics is Bayesian statistics, for which http://www.inference.phy.cam.ac.uk/itila/book.html is absolutely terrific!

Thanks for the link. I always wanted to work through that book and it is great to see that it is free! Someday I will have the time ...

jason